def load_gird_search(self, model, param_grid):
my_loader = Loader()
train, valid, test = my_loader.load_data('data/train.csv',
'data/test.csv')
X_train, y_train, y_train_registered, y_train_casual = my_loader.create_data(
train, self.input_cols)
self.find_optimal_parameters(X_train, y_train_registered, model,
param_grid)
self.find_optimal_parameters(X_train, y_train_casual, model,
param_grid)
def train(self, path_train, path_test):
my_loader = Loader()
train, valid, _ = my_loader.load_data(path_train, path_test)
X_train, y_train, y_train_registered, y_train_casual = my_loader.create_data(
train, self.input_cols)
X_valid, y_valid, y_valid_registered, y_valid_casual = my_loader.create_data(
valid, self.input_cols)
model_registered = self.gradient_boosting_train(
X_train, y_train_registered)
y_predict_registered = np.exp(
self.model_predict(model_registered, X_valid)) - 1
model_casual = self.gradient_boosting_train(X_train, y_train_casual)
y_predict_casual = np.exp(self.model_predict(model_casual,
X_valid)) - 1
y_predict_count = np.round(y_predict_registered + y_predict_casual)
rmsle = self.get_rmsle(
y_predict_count,
np.exp(y_valid_registered) + np.exp(y_valid_casual) - 2)
print(rmsle)
self.predict(path_test, model_registered, model_casual)
